[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"portal-settings:stajic:sr":3,"public-menus:all":38,"post:why-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer:sr":221,"related:post:why-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer:sr:1":1331},{"statusCode":4,"data":5,"message":37},200,{"tenantId":6,"lang":7,"defaultLang":8,"siteUrl":9,"contactEmail":10,"brandName":11,"logoUrl":12,"siteName":11,"siteDescription":13,"ogImage":10,"robotsIndex":14,"socialLinks":10,"reservedSlugs":10,"seoPolicy":15},"stajic","sr","de","https:\u002F\u002Fstajic.de",null,"Stajic Platform","\u002FLogo_Planet.svg","Stajic Portal",true,{"branding":16,"relatedContent":17,"crossDomainLinks":18},{"logoUrl":12},{"enabled":14},[19,22,25,28,31,34],{"url":20,"label":21,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Ffigure.rocks","figure.rocks",{"url":23,"label":24,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Floving.rocks","loving.rocks",{"url":26,"label":27,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.com","bazify.com",{"url":29,"label":30,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.de","bazify.de",{"url":32,"label":33,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.at","bazify.at",{"url":35,"label":36,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.ba","bazify.ba","Portal settings resolved",[39,45],{"id":40,"name":41,"location":42,"isActive":14,"isDefault":43,"items":44},1,"main-navigation","header",false,[],{"id":46,"name":47,"location":48,"isActive":14,"isDefault":14,"items":49},4,"main-menu","sidebar",[50,66,79,93,103,118,133],{"id":51,"title":52,"url":60,"target":61,"icon":62,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":64,"portfolioId":10,"children":65},"item-18",{"de":53,"en":54,"es":55,"fr":56,"it":54,"ru":57,"sr":58,"zh":59},"Startseite","Home","Inicio","Accueil","Главная","Почетна","首页","\u002Ffull-stack-web-developer-munich-performance-seo-and-maintainable-builds","_self","i-lucide-home","page",111,[],{"id":67,"title":68,"url":75,"target":61,"icon":76,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":77,"portfolioId":10,"children":78},"item-22",{"de":69,"en":69,"es":70,"fr":69,"it":71,"ru":72,"sr":73,"zh":74},"Vision","Visión","Visione","Видение","Визија","想象","\u002Fueber-uns-webdesign-muenchen-webaplikation","i-lucide-eye",113,[],{"id":80,"title":81,"url":89,"target":61,"icon":90,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":91,"portfolioId":10,"children":92},"item-19",{"de":82,"en":83,"es":84,"fr":83,"it":85,"ru":86,"sr":87,"zh":88},"Leistungen","Services","Servicios","Servizi","Услуги","Услуге","服务","\u002Fservices-dienstleistungen-muenchen","i-lucide-wrench",116,[],{"id":94,"title":95,"url":99,"target":61,"icon":100,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":101,"portfolioId":10,"children":102},"item-23",{"de":96,"en":96,"es":96,"fr":96,"it":96,"ru":97,"sr":97,"zh":98},"Blog","Блог","博客","\u002Fblog","i-lucide-book-open",112,[],{"id":104,"title":105,"url":114,"target":61,"icon":115,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":116,"portfolioId":10,"children":117},"item-32",{"de":106,"en":107,"es":108,"fr":109,"it":110,"ru":111,"sr":112,"zh":113},"Neue Technologien","New Technologies","Nuevas tecnologías","Nouvelles technologies","Nuove tecnologie","Новые технологии","Нове технологије","新技术！","\u002Fneue-webtechnologien","i-lucide-sparkles",122,[],{"id":119,"title":120,"url":129,"target":61,"icon":130,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":131,"portfolioId":10,"children":132},"item-20",{"de":121,"en":122,"es":123,"fr":124,"it":125,"ru":126,"sr":127,"zh":128},"Kontakt","Contact us!","Contacto","Contact","Contatto","Контакт","Контактирајте нас","联系我们！","\u002Fcontact","i-lucide-mail",115,[],{"id":134,"title":135,"url":144,"target":61,"icon":145,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":146,"portfolioId":10,"children":147},"item-21",{"de":136,"en":137,"es":138,"fr":139,"it":140,"ru":141,"sr":142,"zh":143},"Unsere Arbeit","Our Work","Nuestro trabajo","Nos réalisations","I nostri lavori","Наши работы","Наши радови","文件夹","\u002Fportfolio","i-lucide-briefcase",114,[148,161,175,191,206],{"id":149,"title":150,"url":144,"target":61,"icon":159,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":146,"portfolioId":10,"children":160},"item-24",{"de":151,"en":152,"es":153,"fr":154,"it":155,"ru":156,"sr":157,"zh":158},"Alle Projekte","All Projects","Todos los proyectos","Tous les projets","Tutti i progetti","Все проекты","Сви пројекти","所有项目","i-lucide-grid-3x3",[],{"id":162,"title":163,"url":171,"target":61,"icon":172,"isActive":14,"type":173,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":174},"item-29",{"de":164,"en":165,"es":166,"fr":167,"it":168,"ru":169,"sr":170,"zh":143},"Local Roots, Global Reach","Local Roots - Global Reach","Empresa local ","Entreprise locale","Azienda locale","Местная компания","Локално предузеће глобално тржиште","\u002Fportfolio\u002Flocal-roots-global-reach-communication-media-systems-for-modern-business","i-lucide-globe-2","custom",[],{"id":176,"title":177,"url":186,"target":61,"icon":187,"isActive":14,"type":188,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":189,"children":190},"menu-2916d060-8681-482a-af9a-c107a8bc30d8",{"de":178,"en":179,"es":180,"fr":181,"it":182,"ru":183,"sr":184,"zh":185},"Von Rohdaten zu verifizierbarem KI-Kontext: Eine praktische Architektur für agentenbereite Informationen","From Raw Data to Verifiable AI Context:","De datos sin procesar a contexto de IA verificable: una arquitectura práctica para información lista para agentes","Des données brutes au contexte d'IA vérifiable : une architecture pratique pour des informations prêtes pour les agents","Dai dati grezzi a un contesto IA verificabile: Un'architettura pratica per informazioni pronte per gli agenti","От необработанных данных к верифицируемому контексту ИИ: практическая архитектура информации, готовой для агентов","Od sirovih podataka do proverljivog AI konteksta: Praktična arhitektura za informacije spremne za agente","从原始数据到可验证的 AI 上下文：面向智能体就绪信息的实用架构","\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information","i-lucide-database","portfolio",22,[],{"id":192,"title":193,"url":202,"target":61,"icon":203,"isActive":14,"type":188,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":204,"children":205},"menu-e2061cd7-ff53-49d5-a0e8-3d49090577c0",{"de":194,"en":195,"es":196,"fr":197,"it":198,"ru":199,"sr":200,"zh":201},"Präsenzplattformen — Unternehmenskommunikation","Presence Platforms — Enterprise Communication","Plataformas de presencia — Comunicación empresarial","Plateformes de présence — Communication d'entreprise","Presence Platforms — Comunicazione aziendale","Платформы присутствия — Корпоративные коммуникации","Presence platforme — Poslovna komunikacija","Presence 平台 — 企业通信","\u002Fportfolio\u002Fpresence-platforms-enterprise-communication-that-wins-without-competing","i-lucide-badge-percent",21,[],{"id":207,"title":208,"url":217,"target":61,"icon":218,"isActive":14,"type":188,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":219,"children":220},"menu-a781582e-e90d-418e-b8c9-b070bc37c37e",{"de":209,"en":210,"es":211,"fr":212,"it":213,"ru":214,"sr":215,"zh":216},"Maximierung des POS-ROI durch Konsistenz","Maximizing POS ROI Through Consistent","Maximizar el ROI del POS a través de constante","Maximiser le ROI du PDV grâce à la cohérence","Massimizzare il ROI del POS attraverso costante","Максимизация ROI POS за счет последовательного","Maksimizacija POS ROI-ja kroz dosledno","通过持续最大化 POS 投资回报率","\u002Fportfolio\u002Ffrom-vision-to-value-maximizing-pos-roi-through-consistent-ci-creation-and-strategic-branding","i-lucide-briefcase-business",11,[],{"statusCode":4,"data":222,"message":1330},{"id":223,"title":224,"slug":225,"content":226,"contentJson":227,"excerpt":683,"featuredImage":684,"featuredImageAlt":685,"featuredImageCaption":10,"featuredImageTitle":10,"featuredImageCopyright":10,"featuredImageAuthor":10,"featuredImageSourceUrl":10,"featuredImageLicense":10,"featuredImageIsAiGenerated":43,"status":686,"publishedAt":687,"createdAt":688,"updatedAt":689,"seoLocalePaths":690,"categories":699,"author":708,"translations":712},"496","Zašto AI pronalazi neke izvore, a ignoriše druge? — Sloj autoriteta za pronalaženje","why-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u003Ch2 id=\"section-1\">Pitanje\u003C\u002Fh2>\n\u003Cp>Zašto AI sistem za pronalaženje bira jedan izvor dok ignoriše drugi izvor koji izgleda da govori o potpuno istoj temi?\u003C\u002Fp>\n\u003Cp>Semantička relevantnost je samo deo odgovora. Dva dokumenta mogu biti podjednako slična upitu, a istovremeno se radikalno razlikovati po autoritetu, svežini, poreklu, obimu, dozvolama i primenljivosti.\u003C\u002Fp>\n\u003Cp>Ovaj članak koristi termin Sloj autoriteta pronalaženja za granicu selekcije izvora koja se nalazi između pronalaženja potencijalno relevantnih informacija i tretiranja tih informacija kao odgovarajućeg dokaza.\u003C\u002Fp>\n\u003Cblockquote class=\"border-l-4 border-gray-300 pl-4 italic\">Sloj autoriteta pronalaženja je skup pravila za selekciju izvora i svojstava izvora koji određuju koje informacije koje se mogu pronaći zaslužuju da postanu dokaz za određeno pitanje.\u003Ccite class=\"block mt-2 text-sm\">— Radna definicija — sistemski koncept koji se koristi u ovom članku, a ne standardizovani istraživački termin.\u003C\u002Fcite>\u003C\u002Fblockquote>\n\u003Cnav class=\"editorjs-toc\" data-editorjs-toc=\"true\" aria-label=\"Sadržaj\">\u003Cstrong class=\"editorjs-toc__title\">Sadržaj\u003C\u002Fstrong>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-0\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-1\" class=\"editorjs-toc__link\">Pitanje\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-7\" class=\"editorjs-toc__link\">Šta ovo zaista znači\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-13\" class=\"editorjs-toc__link\">Najjednostavniji primer\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-19\" class=\"editorjs-toc__link\">Gde primer prestaje da funkcioniše\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-23\" class=\"editorjs-toc__link\">Direktan odgovor\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-27\" class=\"editorjs-toc__link\">Zašto sličnost sama po sebi nije dovoljna\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-31\" class=\"editorjs-toc__link\">Šest signala autoriteta\u003C\u002Fa>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-1\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-32\" class=\"editorjs-toc__link\">1. Identitet izvora\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-34\" class=\"editorjs-toc__link\">2. Poreklo\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-36\" class=\"editorjs-toc__link\">3. Svežina\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-38\" class=\"editorjs-toc__link\">4. Primenljivost\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-40\" class=\"editorjs-toc__link\">5. Specifičnost dokaza\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-42\" class=\"editorjs-toc__link\">6. Legitimnost pristupa\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-45\" class=\"editorjs-toc__link\">Minimalni model ovlašćenja\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-49\" class=\"editorjs-toc__link\">Veza sa informacijama spremnim za agente\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-53\" class=\"editorjs-toc__link\">Dijagnostička \u002F odlučujuća metoda\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-57\" class=\"editorjs-toc__link\">Uobičajeni načini neuspeha\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-62\" class=\"editorjs-toc__link\">Rubni slučaj: Sukobljeni merodavni izvori\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-66\" class=\"editorjs-toc__link\">Ograničenja\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-70\" class=\"editorjs-toc__link\">Šta bi promenilo ovaj odgovor?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-73\" class=\"editorjs-toc__link\">Zaključak\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-78\" class=\"editorjs-toc__link\">Primarni izvori\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fnav>\n\u003Ch2 id=\"section-7\">Šta ovo zaista znači\u003C\u002Fh2>\n\u003Cp>Sistem za pronalaženje odgovara na relativno usko pitanje: Koje stavke izgledaju relevantno za ovaj upit?\u003C\u002Fp>\n\u003Cp>Produkcijski AI sistem obično treba da odgovori na teže pitanje: Kojim od tih relevantnih stavki treba dozvoliti da utiču na odgovor?\u003C\u002Fp>\n\u003Cp>To drugo pitanje uvodi identitet izvora, verziju, autoritet, period važenja, poreklo i obim. Relevantna stara politika, kopirani blog post ili nezvanični rezime mogu se rangirati izuzetno dobro, a ipak biti pogrešan dokaz za odluku.\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>Question\n   ↓\nRetrieval Trigger\n   ↓\nCandidate Sources\n   ↓\nRETRIEVAL AUTHORITY LAYER\n   ↓\nQualified Evidence\n   ↓\nContext \u002F Reasoning\n   ↓\nAnswer Validity Boundary\n   ↓\nAnswer\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ca href=\"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhen-should-an-ai-stop-trusting-its-own-knowledge-the-retrieval-trigger\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">Kada AI treba da prestane da veruje sopstvenom znanju? — Okidač pronalaženja\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Okidač pronalaženja odlučuje kada su potrebni spoljni dokazi. Sloj autoriteta pronalaženja pita koji izvori treba da ih pruže.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ch2 id=\"section-13\">Najjednostavniji primer\u003C\u002Fh2>\n\u003Cp>Zamislite da zaposleni pita poslovnog asistenta: Koliko dana godišnjeg odmora dobijaju zaposleni?\u003C\u002Fp>\n\u003Cdiv class=\"overflow-x-auto\">\u003Ctable class=\"w-full border-collapse\">\u003Cthead>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Kandidat\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Sadržaj\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Semantička relevantnost\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">A\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Trenutna HR politika, odobrena verzija 6.2\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Visoka\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">B\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Stari vodič za uvođenje u posao kopiran iz verzije 4.8\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Visoka\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Cp>Oba odlomka mogu sadržati skoro identičan tekst. Zbog toga ih rezultat sličnosti može smestiti veoma blizu.\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>Semantic relevance\n        ≠\nSource authority\n        ≠\nApplicability\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>Trenutna odobrena HR politika obično treba da dominira jer je ona merodavan izvor za tu tvrdnju. Stariji vodič ostaje relevantan tekst, ali ne treba da nosi istu dokaznu težinu.\u003C\u002Fp>\n\u003Ch2 id=\"section-19\">Gde primer prestaje da funkcioniše\u003C\u002Fh2>\n\u003Cp>Autoritet nije uvek univerzalna hijerarhija u kojoj jedan izvor pobeđuje svaki put.\u003C\u002Fp>\n\u003Cp>Transakciona baza podataka može biti merodavna za trenutni status porudžbine. Uputstvo za proizvod može biti merodavno za predviđeno funkcionisanje. Potpisan ugovor može kontrolisati komercijalnu obavezu. Primarni naučni rad može biti najbolji izvor za sopstveni eksperiment, dok kasniji sistematski pregled može biti jači za širi zaključak.\u003C\u002Fp>\n\u003Cp>Autoritet je stoga relativan u odnosu na tvrdnju. Ispravna klasa izvora zavisi od onoga što se pita.\u003C\u002Fp>\n\u003Ch2 id=\"section-23\">Direktan odgovor\u003C\u002Fh2>\n\u003Cp>Izvor treba da postane preferirani dokaz za pretragu kada nije samo relevantan, već i dovoljno merodavan, aktuelan, primenljiv, sledljiv i dostupan za konkretnu tvrdnju na koju se odgovara.\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>Useful retrieval evidence\n=\nrelevance\n+ authority\n+ provenance\n+ freshness\n+ applicability\n+ accessibility\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>Ova svojstva ne garantuju istinu. Ona stvaraju jaču osnovu za odlučivanje koja pronađena informacija zaslužuje da uđe u proces zaključivanja.\u003C\u002Fp>\n\u003Ch2 id=\"section-27\">Zašto sličnost sama po sebi nije dovoljna\u003C\u002Fh2>\n\u003Cp>Vektorska pretraga je izuzetno korisna jer može pronaći semantički povezane informacije čak i kada se formulacija razlikuje. Ali semantička sličnost sama po sebi ne zna da li je dokument zastareo, zamenjen, nezvaničan, neovlašćen ili važeći za drugu jurisdikciju.\u003C\u002Fp>\n\u003Cp>Originalna RAG arhitektura pokazala je kako eksterna neparametarska memorija može proširiti znanje modela. Produkcioni sistemi moraju dodati kontrole specifične za aplikaciju oko tog procesa pretrage.\u003C\u002Fp>\n\u003Ca href=\"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhere-does-an-llm-get-its-data-rag-data-sources-in-python\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">Odakle LLM dobija svoje podatke? RAG izvori podataka u Python-u\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Datoteke, baze podataka, API-ji, pretraga i vektorski indeksi mogu svi pružiti znanje u vreme izvršavanja. Metod pretrage treba da odgovara izvoru i pitanju.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ch2 id=\"section-31\">Šest signala autoriteta\u003C\u002Fh2>\n\u003Ch3 id=\"section-32\">1. Identitet izvora\u003C\u002Fh3>\n\u003Cp>Sistem treba da zna šta je izvor zapravo: politika, ugovor, API, zapis u bazi podataka, naučni rad, stranica dokumentacije, korisnička datoteka ili komentar treće strane.\u003C\u002Fp>\n\u003Ch3 id=\"section-34\">2. Poreklo\u003C\u002Fh3>\n\u003Cp>Dokaz treba da sačuva odakle potiče, kako je ušao u sistem i koji originalni objekat ili dokument može biti pregledan ako se tvrdnja ospori.\u003C\u002Fp>\n\u003Ch3 id=\"section-36\">3. Svežina\u003C\u002Fh3>\n\u003Cp>Svežina je važna kada se osnovna činjenica promeni. Aktuelni API odgovor i pet godina star članak mogu biti podjednako relevantni za reči upita, ali ne i za njegov vremenski zahtev.\u003C\u002Fp>\n\u003Ch3 id=\"section-38\">4. Primenljivost\u003C\u002Fh3>\n\u003Cp>Činjenica može biti tačna, ali primenljiva na pogrešnu verziju proizvoda, zemlju, zakupca, grupu kupaca, ugovor, izdanje softvera ili vremenski period.\u003C\u002Fp>\n\u003Ch3 id=\"section-40\">5. Specifičnost dokaza\u003C\u002Fh3>\n\u003Cp>Primarni dokaz koji direktno podržava tvrdnju generalno je jači od lanca sažetaka koji je samo ponavlja.\u003C\u002Fp>\n\u003Ch3 id=\"section-42\">6. Legitimnost pristupa\u003C\u002Fh3>\n\u003Cp>Tehnički dostupan dokument nije nužno autorizovani kontekst. Ovlašćenje za pretragu mora da funkcioniše unutar granica zakupca, dozvola, poverljivosti i minimizacije podataka.\u003C\u002Fp>\n\u003Ca href=\"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhat-is-context-engineering-what-the-model-receives-before-it-answers\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">Šta je kontekstualni inženjering? Šta model prima pre nego što odgovori\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Sastavljanje konteksta je takođe granica validnosti i bezbednosti. Preuzete informacije moraju da očuvaju obim, ovlašćenje i ograničenja pristupa.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ch2 id=\"section-45\">Minimalni model ovlašćenja\u003C\u002Fh2>\n\u003Cp>Praktičan sistem može eksplicitno da predstavi ova svojstva umesto da se nada da će rangiranje po sličnosti da ih obuhvati.\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>{\n  "source": "hr-policy-6.2",\n  "sourceType": "approved_policy",\n  "effectiveFrom": "2026-01-01",\n  "supersedes": "hr-policy-5.9",\n  "jurisdiction": "DE",\n  "authority": "primary",\n  "accessScope": "employee",\n  "provenance": "hr-policy-system"\n}\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>Pretraživač i dalje može da koristi semantičku ili leksičku relevantnost. Sloj ovlašćenja dodaje ograničenja i signale za ponovno rangiranje koje sama relevantnost ne može da pruži.\u003C\u002Fp>\n\u003Ch2 id=\"section-49\">Veza sa informacijama spremnim za agente\u003C\u002Fh2>\n\u003Cp>Ovo postaje posebno važno kada potrošač nije čovek koji čita dokumente, već AI agent koji donosi odluke.\u003C\u002Fp>\n\u003Cp>Informacioni objekat spreman za agente treba da očuva identitet, poreklo, konflikte, nepoznanice i uslove validnosti pre faze pretrage. U suprotnom, sistem pokušava da rekonstruiše ovlašćenje nakon što su informacije već spljoštene u anonimni tekst.\u003C\u002Fp>\n\u003Ca href=\"https:\u002F\u002Fstajic.de\u002Fsr\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">Od sirovih podataka do proverljivog AI konteksta: praktična arhitektura za informacije spremne za agente\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Portfolio arhitektura za pretvaranje heterogenih podataka u kontrolisane informacione objekte sa identitetom, poreklom, konfliktima i validnošću.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ch2 id=\"section-53\">Dijagnostička \u002F odlučujuća metoda\u003C\u002Fh2>\n\u003Cp>Za svaku visokovrednu preuzetu tvrdnju, postavite šest pitanja:\u003C\u002Fp>\n\u003Col>\u003Cli>Koji je tačan izvor proizveo ovaj dokaz?\u003C\u002Fli>\u003Cli>Da li je ovaj izvor merodavan za ovu vrstu tvrdnje?\u003C\u002Fli>\u003Cli>Da li su informacije još uvek aktuelne?\u003C\u002Fli>\u003Cli>Da li se primenjuju na ovog korisnika, verziju, jurisdikciju i vremenski period?\u003C\u002Fli>\u003Cli>Može li se tvrdnja pratiti unazad do originalnog dokaza?\u003C\u002Fli>\u003Cli>Da li je korisnik ili agent bio ovlašćen da ga preuzme?\u003C\u002Fli>\u003C\u002Fol>\n\u003Cp>Ako sistem ne može da odgovori na ova pitanja, problem više nije samo kvalitet pretrage. To je problem upravljanja dokazima.\u003C\u002Fp>\n\u003Ch2 id=\"section-57\">Uobičajeni načini neuspeha\u003C\u002Fh2>\n\u003Cp>Najviši rezultat sličnosti pobeđuje. To meša semantičku blizinu sa dokaznim autoritetom.\u003C\u002Fp>\n\u003Cp>Najnoviji izvor uvek pobeđuje. Noviji komentar može biti manje merodavan od starijeg vladajućeg ugovora ili standarda.\u003C\u002Fp>\n\u003Cp>Zvanično znači univerzalno primenljivo. Zvanični izvor i dalje može pripadati drugoj jurisdikciji, verziji ili obimu.\u003C\u002Fp>\n\u003Cp>Javno znači bezbedno za preuzimanje. Informacije mogu biti javne, ali neprikladne za zadatak, dok privatne informacije mogu biti legitimne samo unutar ovlašćenog konteksta.\u003C\u002Fp>\n\u003Ch2 id=\"section-62\">Rubni slučaj: Sukobljeni merodavni izvori\u003C\u002Fh2>\n\u003Cp>Težak slučaj nije merodavan izvor protiv očigledno slabog izvora. To su dva naizgled merodavna izvora koja se ne slažu.\u003C\u002Fp>\n\u003Cp>U tom trenutku, sloj autoriteta treba da sačuva sukob umesto da ga tiho svede na jedan odgovor.\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>Retrieve\n   ↓\nQualify source authority\n   ↓\nDetect conflict\n   ↓\nPreserve disagreement\n   ↓\nReason about scope \u002F version \u002F provenance\n   ↓\nAnswer only inside the supported boundary\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ch2 id=\"section-66\">Ograničenja\u003C\u002Fh2>\n\u003Cp>Sloj autoriteta za pretragu je konceptualna arhitektura, a ne univerzalni algoritam rangiranja.\u003C\u002Fp>\n\u003Cp>Različiti domeni zahtevaju različite modele autoriteta. Pravni, naučni, komercijalni, operativni i lični podaci ne mogu svi koristiti istu hijerarhiju izvora.\u003C\u002Fp>\n\u003Cp>Takođe ne može garantovati da će eksterni sistem pretrage ili AI sistem rangirati stranicu izdavača. Eksterni sistemi rangiranja ostaju van kontrole izdavača.\u003C\u002Fp>\n\u003Ch2 id=\"section-70\">Šta bi promenilo ovaj odgovor?\u003C\u002Fh2>\n\u003Cp>Sistemi pretrage mogu sve više automatski da uče pouzdanost izvora, kombinuju strukturisano poreklo sa neuronskim rangiranjem i pregovaraju o pristupu kroz agentske protokole.\u003C\u002Fp>\n\u003Cp>To može promeniti način na koji se implementiraju signali autoriteta. To ne uklanja razliku između pronalaženja relevantnih informacija i odlučivanja da li te informacije zaslužuju dokaznu težinu.\u003C\u002Fp>\n\u003Ch2 id=\"section-73\">Zaključak\u003C\u002Fh2>\n\u003Cp>Okidač za preuzimanje odgovara na pitanje: Da li su mi potrebni spoljni dokazi?\u003C\u002Fp>\n\u003Cp>Sloj autoriteta za preuzimanje odgovara na sledeće pitanje: Koji spoljni izvor zaslužuje da postane dokaz?\u003C\u002Fp>\n\u003Cpre class=\"code-block\">\u003Ccode>Model Knowledge\n   ↓\nRetrieval Trigger\n   ↓\nCandidate Sources\n   ↓\nRetrieval Authority Layer\n   ↓\nQualified Evidence\n   ↓\nReasoning\n   ↓\nAnswer\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>Ta razlika pomera RAG od preuzimanja dokumenata ka kontrolisanom preuzimanju dokaza.\u003C\u002Fp>\n\u003Ch2 id=\"section-78\">Primarni izvori\u003C\u002Fh2>\n\u003Cp>Patrick Lewis et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401\" target=\"_blank\">Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks\u003C\u002Fa> (2020).\u003C\u002Fp>\n\u003Cp>Akari Asai et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.11511\" target=\"_blank\">Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection\u003C\u002Fa> (2023).\u003C\u002Fp>\n\u003Cp>Soyeong Jeong et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.14403\" target=\"_blank\">Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity\u003C\u002Fa> (2024).\u003C\u002Fp>",{"time":228,"blocks":229,"version":682},1792072530197,[230,236,242,247,252,260,267,272,277,282,287,293,304,309,314,330,335,340,345,350,355,360,365,370,375,380,385,390,395,400,409,414,419,424,429,434,439,444,449,454,459,464,469,474,483,488,493,498,503,508,513,518,527,532,537,552,557,562,567,572,577,582,587,592,597,602,607,612,617,622,627,632,637,642,647,652,657,662,667,672,677],{"id":231,"data":232,"type":42,"tunes":235},"XpcNQW15-B",{"text":233,"level":234},"Pitanje",2,{},{"id":237,"data":238,"type":240,"tunes":241},"ren2ISkSca",{"text":239},"Zašto AI sistem za pronalaženje bira jedan izvor dok ignoriše drugi izvor koji izgleda da govori o potpuno istoj temi?","paragraph",{},{"id":243,"data":244,"type":240,"tunes":246},"X_bzzHXG4I",{"text":245},"Semantička relevantnost je samo deo odgovora. Dva dokumenta mogu biti podjednako slična upitu, a istovremeno se radikalno razlikovati po autoritetu, svežini, poreklu, obimu, dozvolama i primenljivosti.",{},{"id":248,"data":249,"type":240,"tunes":251},"elm6ePGXbF",{"text":250},"Ovaj članak koristi termin Sloj autoriteta pronalaženja za granicu selekcije izvora koja se nalazi između pronalaženja potencijalno relevantnih informacija i tretiranja tih informacija kao odgovarajućeg dokaza.",{},{"id":253,"data":254,"type":258,"tunes":259},"TeJ97szQOn",{"text":255,"caption":256,"alignment":257},"Sloj autoriteta pronalaženja je skup pravila za selekciju izvora i svojstava izvora koji određuju koje informacije koje se mogu pronaći zaslužuju da postanu dokaz za određeno pitanje.","Radna definicija — sistemski koncept koji se koristi u ovom članku, a ne standardizovani istraživački termin.","left","quote",{},{"id":261,"data":262,"type":265,"tunes":266},"JfN6zo2EAW",{"title":263,"maxLevel":264,"minLevel":234},"Sadržaj",3,"tableOfContents",{},{"id":268,"data":269,"type":42,"tunes":271},"RoTT8hQShH",{"text":270,"level":234},"Šta ovo zaista znači",{},{"id":273,"data":274,"type":240,"tunes":276},"Cns53IeLiV",{"text":275},"Sistem za pronalaženje odgovara na relativno usko pitanje: Koje stavke izgledaju relevantno za ovaj upit?",{},{"id":278,"data":279,"type":240,"tunes":281},"2qvTGNQESy",{"text":280},"Produkcijski AI sistem obično treba da odgovori na teže pitanje: Kojim od tih relevantnih stavki treba dozvoliti da utiču na odgovor?",{},{"id":283,"data":284,"type":240,"tunes":286},"KneZrq8Dgr",{"text":285},"To drugo pitanje uvodi identitet izvora, verziju, autoritet, period važenja, poreklo i obim. Relevantna stara politika, kopirani blog post ili nezvanični rezime mogu se rangirati izuzetno dobro, a ipak biti pogrešan dokaz za odluku.",{},{"id":288,"data":289,"type":291,"tunes":292},"DYVDznB9AW",{"code":290},"Question\n   ↓\nRetrieval Trigger\n   ↓\nCandidate Sources\n   ↓\nRETRIEVAL AUTHORITY LAYER\n   ↓\nQualified Evidence\n   ↓\nContext \u002F Reasoning\n   ↓\nAnswer Validity Boundary\n   ↓\nAnswer","code",{},{"id":294,"data":295,"type":302,"tunes":303},"mBPGlmAz5D",{"link":296,"meta":297},"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhen-should-an-ai-stop-trusting-its-own-knowledge-the-retrieval-trigger",{"image":298,"title":300,"description":301},{"url":299},"","Kada AI treba da prestane da veruje sopstvenom znanju? — Okidač pronalaženja","Okidač pronalaženja odlučuje kada su potrebni spoljni dokazi. Sloj autoriteta pronalaženja pita koji izvori treba da ih pruže.","linkTool",{},{"id":305,"data":306,"type":42,"tunes":308},"zuuDXruzRL",{"text":307,"level":234},"Najjednostavniji primer",{},{"id":310,"data":311,"type":240,"tunes":313},"QP0qUL8Ciw",{"text":312},"Zamislite da zaposleni pita poslovnog asistenta: Koliko dana godišnjeg odmora dobijaju zaposleni?",{},{"id":315,"data":316,"type":328,"tunes":329},"2Ev2t8opWr",{"content":317,"stretched":43,"withHeadings":14},[318,321,325],[319,263,320],"Kandidat","Semantička relevantnost",[322,323,324],"A","Trenutna HR politika, odobrena verzija 6.2","Visoka",[326,327,324],"B","Stari vodič za uvođenje u posao kopiran iz verzije 4.8","table",{},{"id":331,"data":332,"type":240,"tunes":334},"wtPrU1Dnqd",{"text":333},"Oba odlomka mogu sadržati skoro identičan tekst. Zbog toga ih rezultat sličnosti može smestiti veoma blizu.",{},{"id":336,"data":337,"type":291,"tunes":339},"v97zTkFBzt",{"code":338},"Semantic relevance\n        ≠\nSource authority\n        ≠\nApplicability",{},{"id":341,"data":342,"type":240,"tunes":344},"kAKa32Hh_B",{"text":343},"Trenutna odobrena HR politika obično treba da dominira jer je ona merodavan izvor za tu tvrdnju. Stariji vodič ostaje relevantan tekst, ali ne treba da nosi istu dokaznu težinu.",{},{"id":346,"data":347,"type":42,"tunes":349},"gUE1_pfNG9",{"text":348,"level":234},"Gde primer prestaje da funkcioniše",{},{"id":351,"data":352,"type":240,"tunes":354},"3c0r-JJC4s",{"text":353},"Autoritet nije uvek univerzalna hijerarhija u kojoj jedan izvor pobeđuje svaki put.",{},{"id":356,"data":357,"type":240,"tunes":359},"q1E1rsEJq0",{"text":358},"Transakciona baza podataka može biti merodavna za trenutni status porudžbine. Uputstvo za proizvod može biti merodavno za predviđeno funkcionisanje. Potpisan ugovor može kontrolisati komercijalnu obavezu. Primarni naučni rad može biti najbolji izvor za sopstveni eksperiment, dok kasniji sistematski pregled može biti jači za širi zaključak.",{},{"id":361,"data":362,"type":240,"tunes":364},"oCUaHjt58Z",{"text":363},"Autoritet je stoga relativan u odnosu na tvrdnju. Ispravna klasa izvora zavisi od onoga što se pita.",{},{"id":366,"data":367,"type":42,"tunes":369},"4CxlnDa9Cc",{"text":368,"level":234},"Direktan odgovor",{},{"id":371,"data":372,"type":240,"tunes":374},"QW5_w33xWI",{"text":373},"Izvor treba da postane preferirani dokaz za pretragu kada nije samo relevantan, već i dovoljno merodavan, aktuelan, primenljiv, sledljiv i dostupan za konkretnu tvrdnju na koju se odgovara.",{},{"id":376,"data":377,"type":291,"tunes":379},"owyjAlbwY4",{"code":378},"Useful retrieval evidence\n=\nrelevance\n+ authority\n+ provenance\n+ freshness\n+ applicability\n+ accessibility",{},{"id":381,"data":382,"type":240,"tunes":384},"IBCawe0zx3",{"text":383},"Ova svojstva ne garantuju istinu. Ona stvaraju jaču osnovu za odlučivanje koja pronađena informacija zaslužuje da uđe u proces zaključivanja.",{},{"id":386,"data":387,"type":42,"tunes":389},"4vZIEUjrdq",{"text":388,"level":234},"Zašto sličnost sama po sebi nije dovoljna",{},{"id":391,"data":392,"type":240,"tunes":394},"1PFBw4V0ni",{"text":393},"Vektorska pretraga je izuzetno korisna jer može pronaći semantički povezane informacije čak i kada se formulacija razlikuje. Ali semantička sličnost sama po sebi ne zna da li je dokument zastareo, zamenjen, nezvaničan, neovlašćen ili važeći za drugu jurisdikciju.",{},{"id":396,"data":397,"type":240,"tunes":399},"vfaNbSsKCP",{"text":398},"Originalna RAG arhitektura pokazala je kako eksterna neparametarska memorija može proširiti znanje modela. Produkcioni sistemi moraju dodati kontrole specifične za aplikaciju oko tog procesa pretrage.",{},{"id":401,"data":402,"type":302,"tunes":408},"RKi9NxoiRb",{"link":403,"meta":404},"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhere-does-an-llm-get-its-data-rag-data-sources-in-python",{"image":405,"title":406,"description":407},{"url":299},"Odakle LLM dobija svoje podatke? RAG izvori podataka u Python-u","Datoteke, baze podataka, API-ji, pretraga i vektorski indeksi mogu svi pružiti znanje u vreme izvršavanja. Metod pretrage treba da odgovara izvoru i pitanju.",{},{"id":410,"data":411,"type":42,"tunes":413},"hULFRV_NCz",{"text":412,"level":234},"Šest signala autoriteta",{},{"id":415,"data":416,"type":42,"tunes":418},"7lJ3RudSld",{"text":417,"level":264},"1. Identitet izvora",{},{"id":420,"data":421,"type":240,"tunes":423},"J2JcDKRmnb",{"text":422},"Sistem treba da zna šta je izvor zapravo: politika, ugovor, API, zapis u bazi podataka, naučni rad, stranica dokumentacije, korisnička datoteka ili komentar treće strane.",{},{"id":425,"data":426,"type":42,"tunes":428},"bKliUrdBTI",{"text":427,"level":264},"2. Poreklo",{},{"id":430,"data":431,"type":240,"tunes":433},"LRJssGYfAY",{"text":432},"Dokaz treba da sačuva odakle potiče, kako je ušao u sistem i koji originalni objekat ili dokument može biti pregledan ako se tvrdnja ospori.",{},{"id":435,"data":436,"type":42,"tunes":438},"N-KXk6Cp2H",{"text":437,"level":264},"3. Svežina",{},{"id":440,"data":441,"type":240,"tunes":443},"FJKVbQVgOr",{"text":442},"Svežina je važna kada se osnovna činjenica promeni. Aktuelni API odgovor i pet godina star članak mogu biti podjednako relevantni za reči upita, ali ne i za njegov vremenski zahtev.",{},{"id":445,"data":446,"type":42,"tunes":448},"8Wxcbhl365",{"text":447,"level":264},"4. Primenljivost",{},{"id":450,"data":451,"type":240,"tunes":453},"8w01u-EYq6",{"text":452},"Činjenica može biti tačna, ali primenljiva na pogrešnu verziju proizvoda, zemlju, zakupca, grupu kupaca, ugovor, izdanje softvera ili vremenski period.",{},{"id":455,"data":456,"type":42,"tunes":458},"NYQm_3y_PX",{"text":457,"level":264},"5. Specifičnost dokaza",{},{"id":460,"data":461,"type":240,"tunes":463},"uqqpEuQdxC",{"text":462},"Primarni dokaz koji direktno podržava tvrdnju generalno je jači od lanca sažetaka koji je samo ponavlja.",{},{"id":465,"data":466,"type":42,"tunes":468},"GggJkQwr1C",{"text":467,"level":264},"6. Legitimnost pristupa",{},{"id":470,"data":471,"type":240,"tunes":473},"spnPYg1KRU",{"text":472},"Tehnički dostupan dokument nije nužno autorizovani kontekst. Ovlašćenje za pretragu mora da funkcioniše unutar granica zakupca, dozvola, poverljivosti i minimizacije podataka.",{},{"id":475,"data":476,"type":302,"tunes":482},"yvkNCwWFiC",{"link":477,"meta":478},"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhat-is-context-engineering-what-the-model-receives-before-it-answers",{"image":479,"title":480,"description":481},{"url":299},"Šta je kontekstualni inženjering? Šta model prima pre nego što odgovori","Sastavljanje konteksta je takođe granica validnosti i bezbednosti. Preuzete informacije moraju da očuvaju obim, ovlašćenje i ograničenja pristupa.",{},{"id":484,"data":485,"type":42,"tunes":487},"b_9T6lmCVJ",{"text":486,"level":234},"Minimalni model ovlašćenja",{},{"id":489,"data":490,"type":240,"tunes":492},"VpHVLljSPR",{"text":491},"Praktičan sistem može eksplicitno da predstavi ova svojstva umesto da se nada da će rangiranje po sličnosti da ih obuhvati.",{},{"id":494,"data":495,"type":291,"tunes":497},"L4W7ZQSTYs",{"code":496},"{\n  \"source\": \"hr-policy-6.2\",\n  \"sourceType\": \"approved_policy\",\n  \"effectiveFrom\": \"2026-01-01\",\n  \"supersedes\": \"hr-policy-5.9\",\n  \"jurisdiction\": \"DE\",\n  \"authority\": \"primary\",\n  \"accessScope\": \"employee\",\n  \"provenance\": \"hr-policy-system\"\n}",{},{"id":499,"data":500,"type":240,"tunes":502},"VISBdGcdtA",{"text":501},"Pretraživač i dalje može da koristi semantičku ili leksičku relevantnost. Sloj ovlašćenja dodaje ograničenja i signale za ponovno rangiranje koje sama relevantnost ne može da pruži.",{},{"id":504,"data":505,"type":42,"tunes":507},"GcbcUHNia6",{"text":506,"level":234},"Veza sa informacijama spremnim za agente",{},{"id":509,"data":510,"type":240,"tunes":512},"CVqLmnn5CD",{"text":511},"Ovo postaje posebno važno kada potrošač nije čovek koji čita dokumente, već AI agent koji donosi odluke.",{},{"id":514,"data":515,"type":240,"tunes":517},"MkvgpsEqJc",{"text":516},"Informacioni objekat spreman za agente treba da očuva identitet, poreklo, konflikte, nepoznanice i uslove validnosti pre faze pretrage. U suprotnom, sistem pokušava da rekonstruiše ovlašćenje nakon što su informacije već spljoštene u anonimni tekst.",{},{"id":519,"data":520,"type":302,"tunes":526},"5KX3sugIpt",{"link":521,"meta":522},"https:\u002F\u002Fstajic.de\u002Fsr\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information",{"image":523,"title":524,"description":525},{"url":299},"Od sirovih podataka do proverljivog AI konteksta: praktična arhitektura za informacije spremne za agente","Portfolio arhitektura za pretvaranje heterogenih podataka u kontrolisane informacione objekte sa identitetom, poreklom, konfliktima i validnošću.",{},{"id":528,"data":529,"type":42,"tunes":531},"O6bYy5LenW",{"text":530,"level":234},"Dijagnostička \u002F odlučujuća metoda",{},{"id":533,"data":534,"type":240,"tunes":536},"cZeNmj3Z_3",{"text":535},"Za svaku visokovrednu preuzetu tvrdnju, postavite šest pitanja:",{},{"id":538,"data":539,"type":550,"tunes":551},"lWP8qNNrxw",{"meta":540,"items":542,"style":549},{"counterType":541},"numeric",[543,544,545,546,547,548],"Koji je tačan izvor proizveo ovaj dokaz?","Da li je ovaj izvor merodavan za ovu vrstu tvrdnje?","Da li su informacije još uvek aktuelne?","Da li se primenjuju na ovog korisnika, verziju, jurisdikciju i vremenski period?","Može li se tvrdnja pratiti unazad do originalnog dokaza?","Da li je korisnik ili agent bio ovlašćen da ga preuzme?","ordered","list",{},{"id":553,"data":554,"type":240,"tunes":556},"zzgllTPRBV",{"text":555},"Ako sistem ne može da odgovori na ova pitanja, problem više nije samo kvalitet pretrage. To je problem upravljanja dokazima.",{},{"id":558,"data":559,"type":42,"tunes":561},"gmdM4uPGYl",{"text":560,"level":234},"Uobičajeni načini neuspeha",{},{"id":563,"data":564,"type":240,"tunes":566},"GjbXINllzx",{"text":565},"Najviši rezultat sličnosti pobeđuje. To meša semantičku blizinu sa dokaznim autoritetom.",{},{"id":568,"data":569,"type":240,"tunes":571},"DapWd1cmb5",{"text":570},"Najnoviji izvor uvek pobeđuje. Noviji komentar može biti manje merodavan od starijeg vladajućeg ugovora ili standarda.",{},{"id":573,"data":574,"type":240,"tunes":576},"VVGLrdBjxS",{"text":575},"Zvanično znači univerzalno primenljivo. Zvanični izvor i dalje može pripadati drugoj jurisdikciji, verziji ili obimu.",{},{"id":578,"data":579,"type":240,"tunes":581},"vBrSvczrCq",{"text":580},"Javno znači bezbedno za preuzimanje. Informacije mogu biti javne, ali neprikladne za zadatak, dok privatne informacije mogu biti legitimne samo unutar ovlašćenog konteksta.",{},{"id":583,"data":584,"type":42,"tunes":586},"gN24OPQDE2",{"text":585,"level":234},"Rubni slučaj: Sukobljeni merodavni izvori",{},{"id":588,"data":589,"type":240,"tunes":591},"qcST_7WVLw",{"text":590},"Težak slučaj nije merodavan izvor protiv očigledno slabog izvora. To su dva naizgled merodavna izvora koja se ne slažu.",{},{"id":593,"data":594,"type":240,"tunes":596},"E9gsL-vSes",{"text":595},"U tom trenutku, sloj autoriteta treba da sačuva sukob umesto da ga tiho svede na jedan odgovor.",{},{"id":598,"data":599,"type":291,"tunes":601},"CBZ5ctc6_X",{"code":600},"Retrieve\n   ↓\nQualify source authority\n   ↓\nDetect conflict\n   ↓\nPreserve disagreement\n   ↓\nReason about scope \u002F version \u002F provenance\n   ↓\nAnswer only inside the supported boundary",{},{"id":603,"data":604,"type":42,"tunes":606},"xeQqLNvjCb",{"text":605,"level":234},"Ograničenja",{},{"id":608,"data":609,"type":240,"tunes":611},"raTFpBhsAR",{"text":610},"Sloj autoriteta za pretragu je konceptualna arhitektura, a ne univerzalni algoritam rangiranja.",{},{"id":613,"data":614,"type":240,"tunes":616},"G6tXMIP-u1",{"text":615},"Različiti domeni zahtevaju različite modele autoriteta. Pravni, naučni, komercijalni, operativni i lični podaci ne mogu svi koristiti istu hijerarhiju izvora.",{},{"id":618,"data":619,"type":240,"tunes":621},"SWCg4MmSIb",{"text":620},"Takođe ne može garantovati da će eksterni sistem pretrage ili AI sistem rangirati stranicu izdavača. Eksterni sistemi rangiranja ostaju van kontrole izdavača.",{},{"id":623,"data":624,"type":42,"tunes":626},"j_xI-ClxxP",{"text":625,"level":234},"Šta bi promenilo ovaj odgovor?",{},{"id":628,"data":629,"type":240,"tunes":631},"xrctSdgP3p",{"text":630},"Sistemi pretrage mogu sve više automatski da uče pouzdanost izvora, kombinuju strukturisano poreklo sa neuronskim rangiranjem i pregovaraju o pristupu kroz agentske protokole.",{},{"id":633,"data":634,"type":240,"tunes":636},"Ka8dOdJHXO",{"text":635},"To može promeniti način na koji se implementiraju signali autoriteta. To ne uklanja razliku između pronalaženja relevantnih informacija i odlučivanja da li te informacije zaslužuju dokaznu težinu.",{},{"id":638,"data":639,"type":42,"tunes":641},"E6SLB3m5CI",{"text":640,"level":234},"Zaključak",{},{"id":643,"data":644,"type":240,"tunes":646},"0_oDIxt-xd",{"text":645},"Okidač za preuzimanje odgovara na pitanje: Da li su mi potrebni spoljni dokazi?",{},{"id":648,"data":649,"type":240,"tunes":651},"mCMupYycYk",{"text":650},"Sloj autoriteta za preuzimanje odgovara na sledeće pitanje: Koji spoljni izvor zaslužuje da postane dokaz?",{},{"id":653,"data":654,"type":291,"tunes":656},"rSohO6bowX",{"code":655},"Model Knowledge\n   ↓\nRetrieval Trigger\n   ↓\nCandidate Sources\n   ↓\nRetrieval Authority Layer\n   ↓\nQualified Evidence\n   ↓\nReasoning\n   ↓\nAnswer",{},{"id":658,"data":659,"type":240,"tunes":661},"f8GMr_eelV",{"text":660},"Ta razlika pomera RAG od preuzimanja dokumenata ka kontrolisanom preuzimanju dokaza.",{},{"id":663,"data":664,"type":42,"tunes":666},"hOWp43J5Dw",{"text":665,"level":234},"Primarni izvori",{},{"id":668,"data":669,"type":240,"tunes":671},"evqsbllTgC",{"text":670},"Patrick Lewis et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401\" target=\"_blank\">Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks\u003C\u002Fa> (2020).",{},{"id":673,"data":674,"type":240,"tunes":676},"gz0XrNndq4",{"text":675},"Akari Asai et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.11511\" target=\"_blank\">Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection\u003C\u002Fa> (2023).",{},{"id":678,"data":679,"type":240,"tunes":681},"NgkPqz5Poz",{"text":680},"Soyeong Jeong et al., \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.14403\" target=\"_blank\">Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity\u003C\u002Fa> (2024).",{},"2.31","Pronalaženje relevantnog dokumenta nije dovoljno. AI sistemi takođe treba da razlikuju izvore koji su aktuelni, merodavni, primenljivi i sledljivi od izvora koji samo izgledaju slično. Sloj autoriteta pronalaženja definiše tu granicu odabira izvora.","\u002Fuploads\u002F2026\u002F10\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer-1792072410616-0evp1z.webp","why-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer-1792072410616-0evp1z","PUBLISHED","2026-10-15T07:51:00.000Z","2026-10-15T13:51:04.579Z","2026-10-15T20:37:39.839Z",{"en":691,"de":692,"sr":693,"es":694,"fr":695,"it":696,"ru":697,"zh":698},"\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fde\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fsr\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fes\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Ffr\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fit\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fru\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer","\u002Fzh\u002Fblog\u002Fwhy-does-ai-retrieve-some-sources-and-ignore-others-the-retrieval-authority-layer",[700,704],{"id":701,"name":702,"slug":703},65,"Sistem tematskog autoriteta","topical-authority",{"id":705,"name":706,"slug":707},106,"Runbook: AI rollback","ai-rollback",{"id":709,"login":710,"email":299,"displayName":711},"20","rooth8233","Aleksandar Stajić",[713,1070],{"lang":714,"title":715,"content":716,"contentJson":717,"excerpt":1069},"en","Why Does AI Retrieve Some Sources and Ignore Others? — The Retrieval Authority Layer","{\"time\":1792072370632,\"blocks\":[{\"id\":\"XpcNQW15-B\",\"data\":{\"text\":\"Question\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"ren2ISkSca\",\"data\":{\"text\":\"Why does an AI retrieval system choose one source while ignoring another source that appears to discuss exactly the same subject?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"X_bzzHXG4I\",\"data\":{\"text\":\"Semantic relevance is only part of the answer. Two documents can be equally similar to a query while differing radically in authority, freshness, provenance, scope, permissions and applicability.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"elm6ePGXbF\",\"data\":{\"text\":\"This article uses the term Retrieval Authority Layer for the source-selection boundary that sits between finding potentially relevant information and treating that information as suitable evidence.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"TeJ97szQOn\",\"data\":{\"text\":\"A Retrieval Authority Layer is the set of source-selection rules and source properties that determine which retrievable information deserves to become evidence for a particular question.\",\"caption\":\"Working definition — a systems concept used in this article, not a standardized research term.\",\"alignment\":\"left\"},\"type\":\"quote\",\"tunes\":{}},{\"id\":\"JfN6zo2EAW\",\"data\":{\"title\":\"Contents\",\"maxLevel\":3,\"minLevel\":2},\"type\":\"tableOfContents\",\"tunes\":{}},{\"id\":\"RoTT8hQShH\",\"data\":{\"text\":\"What This Really Means\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"Cns53IeLiV\",\"data\":{\"text\":\"A retriever answers a relatively narrow question: Which items appear relevant to this query?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"2qvTGNQESy\",\"data\":{\"text\":\"A production AI system usually needs to answer a harder one: Which of those relevant items should be allowed to influence the answer?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"KneZrq8Dgr\",\"data\":{\"text\":\"That second question introduces source identity, version, authority, validity period, provenance and scope. A relevant old policy, copied blog post or unofficial summary can rank extremely well while still being the wrong evidence for the decision.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"DYVDznB9AW\",\"data\":{\"code\":\"Question\\n   ↓\\nRetrieval Trigger\\n   ↓\\nCandidate Sources\\n   ↓\\nRETRIEVAL AUTHORITY LAYER\\n   ↓\\nQualified Evidence\\n   ↓\\nContext \u002F Reasoning\\n   ↓\\nAnswer Validity Boundary\\n   ↓\\nAnswer\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"mBPGlmAz5D\",\"data\":{\"link\":\"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhen-should-an-ai-stop-trusting-its-own-knowledge-the-retrieval-trigger\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"When Should an AI Stop Trusting Its Own Knowledge? — The Retrieval Trigger\",\"description\":\"The Retrieval Trigger decides when external evidence is required. The Retrieval Authority Layer asks which sources should provide it.\"}},\"type\":\"linkTool\",\"tunes\":{}},{\"id\":\"zuuDXruzRL\",\"data\":{\"text\":\"Simplest Example\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"QP0qUL8Ciw\",\"data\":{\"text\":\"Imagine an employee asks an enterprise assistant: How many vacation days do employees receive?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"2Ev2t8opWr\",\"data\":{\"content\":[[\"Candidate\",\"Content\",\"Semantic relevance\"],[\"A\",\"Current HR policy, approved version 6.2\",\"High\"],[\"B\",\"Old onboarding guide copied from version 4.8\",\"High\"]],\"stretched\":false,\"withHeadings\":true},\"type\":\"table\",\"tunes\":{}},{\"id\":\"wtPrU1Dnqd\",\"data\":{\"text\":\"Both passages can contain nearly identical wording. A similarity score may therefore place them very close together.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"v97zTkFBzt\",\"data\":{\"code\":\"Semantic relevance\\n        ≠\\nSource authority\\n        ≠\\nApplicability\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"kAKa32Hh_B\",\"data\":{\"text\":\"The current approved HR policy should normally dominate because it is the governing source for the claim. The older guide remains relevant text, but it should not carry the same evidential weight.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"gUE1_pfNG9\",\"data\":{\"text\":\"Where the Example Stops Working\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"3c0r-JJC4s\",\"data\":{\"text\":\"Authority is not always a universal hierarchy in which one source wins every time.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"q1E1rsEJq0\",\"data\":{\"text\":\"A transactional database may be authoritative for the current status of an order. A product manual may be authoritative for intended operation. A signed contract may control a commercial obligation. A primary scientific paper may be the best source for its experiment, while a later systematic review may be stronger for a broader conclusion.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"oCUaHjt58Z\",\"data\":{\"text\":\"Authority is therefore claim-relative. The correct source class depends on what is being asked.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"4CxlnDa9Cc\",\"data\":{\"text\":\"Direct Answer\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"QW5_w33xWI\",\"data\":{\"text\":\"A source should become preferred retrieval evidence when it is not merely relevant, but also sufficiently authoritative, current, applicable, traceable and accessible for the specific claim being answered.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"owyjAlbwY4\",\"data\":{\"code\":\"Useful retrieval evidence\\n=\\nrelevance\\n+ authority\\n+ provenance\\n+ freshness\\n+ applicability\\n+ accessibility\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"IBCawe0zx3\",\"data\":{\"text\":\"These properties do not guarantee truth. They create a stronger basis for deciding which retrieved information deserves to enter the reasoning process.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"4vZIEUjrdq\",\"data\":{\"text\":\"Why Similarity Alone Is Not Enough\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"1PFBw4V0ni\",\"data\":{\"text\":\"Vector retrieval is extremely useful because it can find semantically related information even when the wording differs. But semantic similarity does not know by itself whether a document is obsolete, superseded, unofficial, unauthorized or valid for another jurisdiction.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"vfaNbSsKCP\",\"data\":{\"text\":\"The original RAG architecture showed how external non-parametric memory can augment model knowledge. Production systems must add application-specific controls around that retrieval process.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"RKi9NxoiRb\",\"data\":{\"link\":\"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhere-does-an-llm-get-its-data-rag-data-sources-in-python\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"Where Does an LLM Get Its Data? RAG Data Sources in Python\",\"description\":\"Files, databases, APIs, search and vector indexes can all provide runtime knowledge. The retrieval method should match the source and the question.\"}},\"type\":\"linkTool\",\"tunes\":{}},{\"id\":\"hULFRV_NCz\",\"data\":{\"text\":\"The Six Authority Signals\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"7lJ3RudSld\",\"data\":{\"text\":\"1. Source Identity\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"J2JcDKRmnb\",\"data\":{\"text\":\"The system should know what the source actually is: policy, contract, API, database record, research paper, documentation page, user file or third-party commentary.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"bKliUrdBTI\",\"data\":{\"text\":\"2. Provenance\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"LRJssGYfAY\",\"data\":{\"text\":\"Evidence should preserve where it came from, how it entered the system and which original object or document can be inspected if the claim is challenged.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"N-KXk6Cp2H\",\"data\":{\"text\":\"3. Freshness\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"FJKVbQVgOr\",\"data\":{\"text\":\"Freshness matters when the underlying fact changes. A current API response and a five-year-old article may be equally relevant to the words of a query but not to its temporal requirement.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"8Wxcbhl365\",\"data\":{\"text\":\"4. Applicability\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"8w01u-EYq6\",\"data\":{\"text\":\"A fact may be correct but applicable to the wrong product version, country, tenant, customer group, contract, software release or time period.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"NYQm_3y_PX\",\"data\":{\"text\":\"5. Evidence Specificity\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"uqqpEuQdxC\",\"data\":{\"text\":\"Primary evidence that directly supports the claim is generally stronger than a chain of summaries that merely repeat it.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"GggJkQwr1C\",\"data\":{\"text\":\"6. Access Legitimacy\",\"level\":3},\"type\":\"header\",\"tunes\":{}},{\"id\":\"spnPYg1KRU\",\"data\":{\"text\":\"A technically retrievable document is not necessarily authorized context. Retrieval authority must operate inside tenant, permission, confidentiality and data-minimization boundaries.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"yvkNCwWFiC\",\"data\":{\"link\":\"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhat-is-context-engineering-what-the-model-receives-before-it-answers\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"What Is Context Engineering? What the Model Receives Before It Answers\",\"description\":\"Context assembly is also a validity and security boundary. Retrieved information must preserve scope, authority and access constraints.\"}},\"type\":\"linkTool\",\"tunes\":{}},{\"id\":\"b_9T6lmCVJ\",\"data\":{\"text\":\"A Minimal Authority Model\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"VpHVLljSPR\",\"data\":{\"text\":\"A practical system can represent these properties explicitly rather than hoping that similarity ranking captures them.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"L4W7ZQSTYs\",\"data\":{\"code\":\"{\\n  \\\"source\\\": \\\"hr-policy-6.2\\\",\\n  \\\"sourceType\\\": \\\"approved_policy\\\",\\n  \\\"effectiveFrom\\\": \\\"2026-01-01\\\",\\n  \\\"supersedes\\\": \\\"hr-policy-5.9\\\",\\n  \\\"jurisdiction\\\": \\\"DE\\\",\\n  \\\"authority\\\": \\\"primary\\\",\\n  \\\"accessScope\\\": \\\"employee\\\",\\n  \\\"provenance\\\": \\\"hr-policy-system\\\"\\n}\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"VISBdGcdtA\",\"data\":{\"text\":\"The retriever can still use semantic or lexical relevance. The authority layer adds constraints and reranking signals that relevance alone cannot provide.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"GcbcUHNia6\",\"data\":{\"text\":\"Connection to Agent-Ready Information\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"CVqLmnn5CD\",\"data\":{\"text\":\"This becomes especially important when the consumer is not a human reading documents but an AI agent making decisions.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"MkvgpsEqJc\",\"data\":{\"text\":\"An agent-ready information object should preserve identity, provenance, conflicts, unknowns and validity conditions before the retrieval stage. Otherwise the system tries to reconstruct authority after information has already been flattened into anonymous text.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"5KX3sugIpt\",\"data\":{\"link\":\"https:\u002F\u002Fstajic.de\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"From Raw Data to Verifiable AI Context: A Practical Architecture for Agent-Ready Information\",\"description\":\"A portfolio architecture for turning heterogeneous data into controlled information objects with identity, provenance, conflicts and validity.\"}},\"type\":\"linkTool\",\"tunes\":{}},{\"id\":\"O6bYy5LenW\",\"data\":{\"text\":\"Diagnostic \u002F Decision Method\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"cZeNmj3Z_3\",\"data\":{\"text\":\"For every high-value retrieved claim, ask six questions:\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"lWP8qNNrxw\",\"data\":{\"meta\":{\"counterType\":\"numeric\"},\"items\":[\"What exact source produced this evidence?\",\"Is this source authoritative for this type of claim?\",\"Is the information still current?\",\"Does it apply to this user, version, jurisdiction and time period?\",\"Can the claim be traced back to the original evidence?\",\"Was the user or agent authorized to retrieve it?\"],\"style\":\"ordered\"},\"type\":\"list\",\"tunes\":{}},{\"id\":\"zzgllTPRBV\",\"data\":{\"text\":\"If the system cannot answer these questions, the problem is no longer only retrieval quality. It is an evidence-governance problem.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"gmdM4uPGYl\",\"data\":{\"text\":\"Common Failure Modes\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"GjbXINllzx\",\"data\":{\"text\":\"The highest similarity score wins. This confuses semantic closeness with evidential authority.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"DapWd1cmb5\",\"data\":{\"text\":\"The newest source always wins. A newer commentary may be less authoritative than an older governing contract or standard.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"VVGLrdBjxS\",\"data\":{\"text\":\"Official means universally applicable. An official source may still belong to another jurisdiction, version or scope.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"vBrSvczrCq\",\"data\":{\"text\":\"Public means safe to retrieve. Information can be public yet inappropriate for the task, while private information can be legitimate only inside an authorized context.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"gN24OPQDE2\",\"data\":{\"text\":\"Edge Case: Conflicting Authoritative Sources\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"qcST_7WVLw\",\"data\":{\"text\":\"The difficult case is not an authoritative source versus an obviously weak source. It is two apparently authoritative sources that disagree.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"E9gsL-vSes\",\"data\":{\"text\":\"At that point, the authority layer should preserve the conflict instead of silently collapsing it into one answer.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"CBZ5ctc6_X\",\"data\":{\"code\":\"Retrieve\\n   ↓\\nQualify source authority\\n   ↓\\nDetect conflict\\n   ↓\\nPreserve disagreement\\n   ↓\\nReason about scope \u002F version \u002F provenance\\n   ↓\\nAnswer only inside the supported boundary\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"xeQqLNvjCb\",\"data\":{\"text\":\"Limitations\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"raTFpBhsAR\",\"data\":{\"text\":\"The Retrieval Authority Layer is a conceptual architecture, not a universal ranking algorithm.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"G6tXMIP-u1\",\"data\":{\"text\":\"Different domains require different authority models. Legal, scientific, commercial, operational and personal data cannot all use the same source hierarchy.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"SWCg4MmSIb\",\"data\":{\"text\":\"It also cannot guarantee that an external search or AI system will rank a publisher's page. External ranking systems remain outside the publisher's control.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"j_xI-ClxxP\",\"data\":{\"text\":\"What Would Change This Answer?\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"xrctSdgP3p\",\"data\":{\"text\":\"Retrieval systems may increasingly learn source reliability automatically, combine structured provenance with neural ranking and negotiate access through agent protocols.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"Ka8dOdJHXO\",\"data\":{\"text\":\"That may change how authority signals are implemented. It does not remove the distinction between finding relevant information and deciding whether that information deserves evidential weight.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"E6SLB3m5CI\",\"data\":{\"text\":\"Conclusion\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"0_oDIxt-xd\",\"data\":{\"text\":\"The Retrieval Trigger answers: Do I need external evidence?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"mCMupYycYk\",\"data\":{\"text\":\"The Retrieval Authority Layer answers the next question: Which external source deserves to become evidence?\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"rSohO6bowX\",\"data\":{\"code\":\"Model Knowledge\\n   ↓\\nRetrieval Trigger\\n   ↓\\nCandidate Sources\\n   ↓\\nRetrieval Authority Layer\\n   ↓\\nQualified Evidence\\n   ↓\\nReasoning\\n   ↓\\nAnswer\"},\"type\":\"code\",\"tunes\":{}},{\"id\":\"f8GMr_eelV\",\"data\":{\"text\":\"That distinction moves RAG from document retrieval toward controlled evidence retrieval.\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"hOWp43J5Dw\",\"data\":{\"text\":\"Primary Sources\",\"level\":2},\"type\":\"header\",\"tunes\":{}},{\"id\":\"evqsbllTgC\",\"data\":{\"text\":\"Patrick Lewis et al., \u003Ca href=\\\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401\\\" target=\\\"_blank\\\">Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks\u003C\u002Fa> (2020).\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"gz0XrNndq4\",\"data\":{\"text\":\"Akari Asai et al., \u003Ca href=\\\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.11511\\\" target=\\\"_blank\\\">Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection\u003C\u002Fa> (2023).\"},\"type\":\"paragraph\",\"tunes\":{}},{\"id\":\"NgkPqz5Poz\",\"data\":{\"text\":\"Soyeong Jeong et al., \u003Ca href=\\\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2403.14403\\\" target=\\\"_blank\\\">Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity\u003C\u002Fa> (2024).\"},\"type\":\"paragraph\",\"tunes\":{}}],\"version\":\"2.31.6\"}",{"time":718,"blocks":719,"version":1068},1792072370632,[720,724,728,732,736,741,745,749,753,757,761,764,772,776,780,793,797,800,804,808,812,816,820,824,828,831,835,839,843,847,855,859,863,867,871,875,879,883,887,891,895,899,903,907,915,919,923,926,930,934,938,942,950,954,958,969,973,977,981,985,989,993,997,1001,1005,1008,1012,1016,1020,1024,1028,1032,1036,1040,1044,1048,1051,1055,1059,1062,1065],{"id":231,"data":721,"type":42,"tunes":723},{"text":722,"level":234},"Question",{},{"id":237,"data":725,"type":240,"tunes":727},{"text":726},"Why does an AI retrieval system choose one source while ignoring another source that appears to discuss exactly the same subject?",{},{"id":243,"data":729,"type":240,"tunes":731},{"text":730},"Semantic relevance is only part of the answer. Two documents can be equally similar to a query while differing radically in authority, freshness, provenance, scope, permissions and applicability.",{},{"id":248,"data":733,"type":240,"tunes":735},{"text":734},"This article uses the term Retrieval Authority Layer for the source-selection boundary that sits between finding potentially relevant information and treating that information as suitable evidence.",{},{"id":253,"data":737,"type":258,"tunes":740},{"text":738,"caption":739,"alignment":257},"A Retrieval Authority Layer is the set of source-selection rules and source properties that determine which retrievable information deserves to become evidence for a particular question.","Working definition — a systems concept used in this article, not a standardized research term.",{},{"id":261,"data":742,"type":265,"tunes":744},{"title":743,"maxLevel":264,"minLevel":234},"Contents",{},{"id":268,"data":746,"type":42,"tunes":748},{"text":747,"level":234},"What This Really Means",{},{"id":273,"data":750,"type":240,"tunes":752},{"text":751},"A retriever answers a relatively narrow question: Which items appear relevant to this query?",{},{"id":278,"data":754,"type":240,"tunes":756},{"text":755},"A production AI system usually needs to answer a harder one: Which of those relevant items should be allowed to influence the answer?",{},{"id":283,"data":758,"type":240,"tunes":760},{"text":759},"That second question introduces source identity, version, authority, validity period, provenance and scope. A relevant old policy, copied blog post or unofficial summary can rank extremely well while still being the wrong evidence for the decision.",{},{"id":288,"data":762,"type":291,"tunes":763},{"code":290},{},{"id":294,"data":765,"type":302,"tunes":771},{"link":766,"meta":767},"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhen-should-an-ai-stop-trusting-its-own-knowledge-the-retrieval-trigger",{"image":768,"title":769,"description":770},{"url":299},"When Should an AI Stop Trusting Its Own Knowledge? — The Retrieval Trigger","The Retrieval Trigger decides when external evidence is required. The Retrieval Authority Layer asks which sources should provide it.",{},{"id":305,"data":773,"type":42,"tunes":775},{"text":774,"level":234},"Simplest Example",{},{"id":310,"data":777,"type":240,"tunes":779},{"text":778},"Imagine an employee asks an enterprise assistant: How many vacation days do employees receive?",{},{"id":315,"data":781,"type":328,"tunes":792},{"content":782,"stretched":43,"withHeadings":14},[783,787,790],[784,785,786],"Candidate","Content","Semantic relevance",[322,788,789],"Current HR policy, approved version 6.2","High",[326,791,789],"Old onboarding guide copied from version 4.8",{},{"id":331,"data":794,"type":240,"tunes":796},{"text":795},"Both passages can contain nearly identical wording. A similarity score may therefore place them very close together.",{},{"id":336,"data":798,"type":291,"tunes":799},{"code":338},{},{"id":341,"data":801,"type":240,"tunes":803},{"text":802},"The current approved HR policy should normally dominate because it is the governing source for the claim. The older guide remains relevant text, but it should not carry the same evidential weight.",{},{"id":346,"data":805,"type":42,"tunes":807},{"text":806,"level":234},"Where the Example Stops Working",{},{"id":351,"data":809,"type":240,"tunes":811},{"text":810},"Authority is not always a universal hierarchy in which one source wins every time.",{},{"id":356,"data":813,"type":240,"tunes":815},{"text":814},"A transactional database may be authoritative for the current status of an order. A product manual may be authoritative for intended operation. A signed contract may control a commercial obligation. A primary scientific paper may be the best source for its experiment, while a later systematic review may be stronger for a broader conclusion.",{},{"id":361,"data":817,"type":240,"tunes":819},{"text":818},"Authority is therefore claim-relative. The correct source class depends on what is being asked.",{},{"id":366,"data":821,"type":42,"tunes":823},{"text":822,"level":234},"Direct Answer",{},{"id":371,"data":825,"type":240,"tunes":827},{"text":826},"A source should become preferred retrieval evidence when it is not merely relevant, but also sufficiently authoritative, current, applicable, traceable and accessible for the specific claim being answered.",{},{"id":376,"data":829,"type":291,"tunes":830},{"code":378},{},{"id":381,"data":832,"type":240,"tunes":834},{"text":833},"These properties do not guarantee truth. They create a stronger basis for deciding which retrieved information deserves to enter the reasoning process.",{},{"id":386,"data":836,"type":42,"tunes":838},{"text":837,"level":234},"Why Similarity Alone Is Not Enough",{},{"id":391,"data":840,"type":240,"tunes":842},{"text":841},"Vector retrieval is extremely useful because it can find semantically related information even when the wording differs. But semantic similarity does not know by itself whether a document is obsolete, superseded, unofficial, unauthorized or valid for another jurisdiction.",{},{"id":396,"data":844,"type":240,"tunes":846},{"text":845},"The original RAG architecture showed how external non-parametric memory can augment model knowledge. Production systems must add application-specific controls around that retrieval process.",{},{"id":401,"data":848,"type":302,"tunes":854},{"link":849,"meta":850},"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhere-does-an-llm-get-its-data-rag-data-sources-in-python",{"image":851,"title":852,"description":853},{"url":299},"Where Does an LLM Get Its Data? RAG Data Sources in Python","Files, databases, APIs, search and vector indexes can all provide runtime knowledge. The retrieval method should match the source and the question.",{},{"id":410,"data":856,"type":42,"tunes":858},{"text":857,"level":234},"The Six Authority Signals",{},{"id":415,"data":860,"type":42,"tunes":862},{"text":861,"level":264},"1. Source Identity",{},{"id":420,"data":864,"type":240,"tunes":866},{"text":865},"The system should know what the source actually is: policy, contract, API, database record, research paper, documentation page, user file or third-party commentary.",{},{"id":425,"data":868,"type":42,"tunes":870},{"text":869,"level":264},"2. Provenance",{},{"id":430,"data":872,"type":240,"tunes":874},{"text":873},"Evidence should preserve where it came from, how it entered the system and which original object or document can be inspected if the claim is challenged.",{},{"id":435,"data":876,"type":42,"tunes":878},{"text":877,"level":264},"3. Freshness",{},{"id":440,"data":880,"type":240,"tunes":882},{"text":881},"Freshness matters when the underlying fact changes. A current API response and a five-year-old article may be equally relevant to the words of a query but not to its temporal requirement.",{},{"id":445,"data":884,"type":42,"tunes":886},{"text":885,"level":264},"4. Applicability",{},{"id":450,"data":888,"type":240,"tunes":890},{"text":889},"A fact may be correct but applicable to the wrong product version, country, tenant, customer group, contract, software release or time period.",{},{"id":455,"data":892,"type":42,"tunes":894},{"text":893,"level":264},"5. Evidence Specificity",{},{"id":460,"data":896,"type":240,"tunes":898},{"text":897},"Primary evidence that directly supports the claim is generally stronger than a chain of summaries that merely repeat it.",{},{"id":465,"data":900,"type":42,"tunes":902},{"text":901,"level":264},"6. Access Legitimacy",{},{"id":470,"data":904,"type":240,"tunes":906},{"text":905},"A technically retrievable document is not necessarily authorized context. Retrieval authority must operate inside tenant, permission, confidentiality and data-minimization boundaries.",{},{"id":475,"data":908,"type":302,"tunes":914},{"link":909,"meta":910},"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhat-is-context-engineering-what-the-model-receives-before-it-answers",{"image":911,"title":912,"description":913},{"url":299},"What Is Context Engineering? What the Model Receives Before It Answers","Context assembly is also a validity and security boundary. Retrieved information must preserve scope, authority and access constraints.",{},{"id":484,"data":916,"type":42,"tunes":918},{"text":917,"level":234},"A Minimal Authority Model",{},{"id":489,"data":920,"type":240,"tunes":922},{"text":921},"A practical system can represent these properties explicitly rather than hoping that similarity ranking captures them.",{},{"id":494,"data":924,"type":291,"tunes":925},{"code":496},{},{"id":499,"data":927,"type":240,"tunes":929},{"text":928},"The retriever can still use semantic or lexical relevance. The authority layer adds constraints and reranking signals that relevance alone cannot provide.",{},{"id":504,"data":931,"type":42,"tunes":933},{"text":932,"level":234},"Connection to Agent-Ready Information",{},{"id":509,"data":935,"type":240,"tunes":937},{"text":936},"This becomes especially important when the consumer is not a human reading documents but an AI agent making decisions.",{},{"id":514,"data":939,"type":240,"tunes":941},{"text":940},"An agent-ready information object should preserve identity, provenance, conflicts, unknowns and validity conditions before the retrieval stage. Otherwise the system tries to reconstruct authority after information has already been flattened into anonymous text.",{},{"id":519,"data":943,"type":302,"tunes":949},{"link":944,"meta":945},"https:\u002F\u002Fstajic.de\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information",{"image":946,"title":947,"description":948},{"url":299},"From Raw Data to Verifiable AI Context: A Practical Architecture for Agent-Ready Information","A portfolio architecture for turning heterogeneous data into controlled information objects with identity, provenance, conflicts and validity.",{},{"id":528,"data":951,"type":42,"tunes":953},{"text":952,"level":234},"Diagnostic \u002F Decision Method",{},{"id":533,"data":955,"type":240,"tunes":957},{"text":956},"For every high-value retrieved claim, ask six questions:",{},{"id":538,"data":959,"type":550,"tunes":968},{"meta":960,"items":961,"style":549},{"counterType":541},[962,963,964,965,966,967],"What exact source produced this evidence?","Is this source authoritative for this type of claim?","Is the information still current?","Does it apply to this user, version, jurisdiction and time period?","Can the claim be traced back to the original evidence?","Was the user or agent authorized to retrieve it?",{},{"id":553,"data":970,"type":240,"tunes":972},{"text":971},"If the system cannot answer these questions, the problem is no longer only retrieval quality. It is an evidence-governance problem.",{},{"id":558,"data":974,"type":42,"tunes":976},{"text":975,"level":234},"Common Failure Modes",{},{"id":563,"data":978,"type":240,"tunes":980},{"text":979},"The highest similarity score wins. This confuses semantic closeness with evidential authority.",{},{"id":568,"data":982,"type":240,"tunes":984},{"text":983},"The newest source always wins. A newer commentary may be less authoritative than an older governing contract or standard.",{},{"id":573,"data":986,"type":240,"tunes":988},{"text":987},"Official means universally applicable. An official source may still belong to another jurisdiction, version or scope.",{},{"id":578,"data":990,"type":240,"tunes":992},{"text":991},"Public means safe to retrieve. Information can be public yet inappropriate for the task, while private information can be legitimate only inside an authorized context.",{},{"id":583,"data":994,"type":42,"tunes":996},{"text":995,"level":234},"Edge Case: Conflicting Authoritative Sources",{},{"id":588,"data":998,"type":240,"tunes":1000},{"text":999},"The difficult case is not an authoritative source versus an obviously weak source. It is two apparently authoritative sources that disagree.",{},{"id":593,"data":1002,"type":240,"tunes":1004},{"text":1003},"At that point, the authority layer should preserve the conflict instead of silently collapsing it into one answer.",{},{"id":598,"data":1006,"type":291,"tunes":1007},{"code":600},{},{"id":603,"data":1009,"type":42,"tunes":1011},{"text":1010,"level":234},"Limitations",{},{"id":608,"data":1013,"type":240,"tunes":1015},{"text":1014},"The Retrieval Authority Layer is a conceptual architecture, not a universal ranking algorithm.",{},{"id":613,"data":1017,"type":240,"tunes":1019},{"text":1018},"Different domains require different authority models. Legal, scientific, commercial, operational and personal data cannot all use the same source hierarchy.",{},{"id":618,"data":1021,"type":240,"tunes":1023},{"text":1022},"It also cannot guarantee that an external search or AI system will rank a publisher's page. External ranking systems remain outside the publisher's control.",{},{"id":623,"data":1025,"type":42,"tunes":1027},{"text":1026,"level":234},"What Would Change This Answer?",{},{"id":628,"data":1029,"type":240,"tunes":1031},{"text":1030},"Retrieval systems may increasingly learn source reliability automatically, combine structured provenance with neural ranking and negotiate access through agent protocols.",{},{"id":633,"data":1033,"type":240,"tunes":1035},{"text":1034},"That may change how authority signals are implemented. It does not remove the distinction between finding relevant information and deciding whether that information deserves evidential weight.",{},{"id":638,"data":1037,"type":42,"tunes":1039},{"text":1038,"level":234},"Conclusion",{},{"id":643,"data":1041,"type":240,"tunes":1043},{"text":1042},"The Retrieval Trigger answers: Do I need external evidence?",{},{"id":648,"data":1045,"type":240,"tunes":1047},{"text":1046},"The Retrieval Authority Layer answers the next question: Which external source deserves to become evidence?",{},{"id":653,"data":1049,"type":291,"tunes":1050},{"code":655},{},{"id":658,"data":1052,"type":240,"tunes":1054},{"text":1053},"That distinction moves RAG from document retrieval toward controlled evidence retrieval.",{},{"id":663,"data":1056,"type":42,"tunes":1058},{"text":1057,"level":234},"Primary Sources",{},{"id":668,"data":1060,"type":240,"tunes":1061},{"text":670},{},{"id":673,"data":1063,"type":240,"tunes":1064},{"text":675},{},{"id":678,"data":1066,"type":240,"tunes":1067},{"text":680},{},"2.31.6","Finding a relevant document is not enough. AI systems also need to distinguish sources that are current, authoritative, applicable and traceable from sources that merely look similar. The Retrieval Authority Layer defines that source-selection boundary.",{"lang":7,"title":224,"content":226,"contentJson":1071,"excerpt":683},{"time":228,"blocks":1072,"version":682},[1073,1076,1079,1082,1085,1088,1091,1094,1097,1100,1103,1106,1111,1114,1117,1124,1127,1130,1133,1136,1139,1142,1145,1148,1151,1154,1157,1160,1163,1166,1171,1174,1177,1180,1183,1186,1189,1192,1195,1198,1201,1204,1207,1210,1215,1218,1221,1224,1227,1230,1233,1236,1241,1244,1247,1252,1255,1258,1261,1264,1267,1270,1273,1276,1279,1282,1285,1288,1291,1294,1297,1300,1303,1306,1309,1312,1315,1318,1321,1324,1327],{"id":231,"data":1074,"type":42,"tunes":1075},{"text":233,"level":234},{},{"id":237,"data":1077,"type":240,"tunes":1078},{"text":239},{},{"id":243,"data":1080,"type":240,"tunes":1081},{"text":245},{},{"id":248,"data":1083,"type":240,"tunes":1084},{"text":250},{},{"id":253,"data":1086,"type":258,"tunes":1087},{"text":255,"caption":256,"alignment":257},{},{"id":261,"data":1089,"type":265,"tunes":1090},{"title":263,"maxLevel":264,"minLevel":234},{},{"id":268,"data":1092,"type":42,"tunes":1093},{"text":270,"level":234},{},{"id":273,"data":1095,"type":240,"tunes":1096},{"text":275},{},{"id":278,"data":1098,"type":240,"tunes":1099},{"text":280},{},{"id":283,"data":1101,"type":240,"tunes":1102},{"text":285},{},{"id":288,"data":1104,"type":291,"tunes":1105},{"code":290},{},{"id":294,"data":1107,"type":302,"tunes":1110},{"link":296,"meta":1108},{"image":1109,"title":300,"description":301},{"url":299},{},{"id":305,"data":1112,"type":42,"tunes":1113},{"text":307,"level":234},{},{"id":310,"data":1115,"type":240,"tunes":1116},{"text":312},{},{"id":315,"data":1118,"type":328,"tunes":1123},{"content":1119,"stretched":43,"withHeadings":14},[1120,1121,1122],[319,263,320],[322,323,324],[326,327,324],{},{"id":331,"data":1125,"type":240,"tunes":1126},{"text":333},{},{"id":336,"data":1128,"type":291,"tunes":1129},{"code":338},{},{"id":341,"data":1131,"type":240,"tunes":1132},{"text":343},{},{"id":346,"data":1134,"type":42,"tunes":1135},{"text":348,"level":234},{},{"id":351,"data":1137,"type":240,"tunes":1138},{"text":353},{},{"id":356,"data":1140,"type":240,"tunes":1141},{"text":358},{},{"id":361,"data":1143,"type":240,"tunes":1144},{"text":363},{},{"id":366,"data":1146,"type":42,"tunes":1147},{"text":368,"level":234},{},{"id":371,"data":1149,"type":240,"tunes":1150},{"text":373},{},{"id":376,"data":1152,"type":291,"tunes":1153},{"code":378},{},{"id":381,"data":1155,"type":240,"tunes":1156},{"text":383},{},{"id":386,"data":1158,"type":42,"tunes":1159},{"text":388,"level":234},{},{"id":391,"data":1161,"type":240,"tunes":1162},{"text":393},{},{"id":396,"data":1164,"type":240,"tunes":1165},{"text":398},{},{"id":401,"data":1167,"type":302,"tunes":1170},{"link":403,"meta":1168},{"image":1169,"title":406,"description":407},{"url":299},{},{"id":410,"data":1172,"type":42,"tunes":1173},{"text":412,"level":234},{},{"id":415,"data":1175,"type":42,"tunes":1176},{"text":417,"level":264},{},{"id":420,"data":1178,"type":240,"tunes":1179},{"text":422},{},{"id":425,"data":1181,"type":42,"tunes":1182},{"text":427,"level":264},{},{"id":430,"data":1184,"type":240,"tunes":1185},{"text":432},{},{"id":435,"data":1187,"type":42,"tunes":1188},{"text":437,"level":264},{},{"id":440,"data":1190,"type":240,"tunes":1191},{"text":442},{},{"id":445,"data":1193,"type":42,"tunes":1194},{"text":447,"level":264},{},{"id":450,"data":1196,"type":240,"tunes":1197},{"text":452},{},{"id":455,"data":1199,"type":42,"tunes":1200},{"text":457,"level":264},{},{"id":460,"data":1202,"type":240,"tunes":1203},{"text":462},{},{"id":465,"data":1205,"type":42,"tunes":1206},{"text":467,"level":264},{},{"id":470,"data":1208,"type":240,"tunes":1209},{"text":472},{},{"id":475,"data":1211,"type":302,"tunes":1214},{"link":477,"meta":1212},{"image":1213,"title":480,"description":481},{"url":299},{},{"id":484,"data":1216,"type":42,"tunes":1217},{"text":486,"level":234},{},{"id":489,"data":1219,"type":240,"tunes":1220},{"text":491},{},{"id":494,"data":1222,"type":291,"tunes":1223},{"code":496},{},{"id":499,"data":1225,"type":240,"tunes":1226},{"text":501},{},{"id":504,"data":1228,"type":42,"tunes":1229},{"text":506,"level":234},{},{"id":509,"data":1231,"type":240,"tunes":1232},{"text":511},{},{"id":514,"data":1234,"type":240,"tunes":1235},{"text":516},{},{"id":519,"data":1237,"type":302,"tunes":1240},{"link":521,"meta":1238},{"image":1239,"title":524,"description":525},{"url":299},{},{"id":528,"data":1242,"type":42,"tunes":1243},{"text":530,"level":234},{},{"id":533,"data":1245,"type":240,"tunes":1246},{"text":535},{},{"id":538,"data":1248,"type":550,"tunes":1251},{"meta":1249,"items":1250,"style":549},{"counterType":541},[543,544,545,546,547,548],{},{"id":553,"data":1253,"type":240,"tunes":1254},{"text":555},{},{"id":558,"data":1256,"type":42,"tunes":1257},{"text":560,"level":234},{},{"id":563,"data":1259,"type":240,"tunes":1260},{"text":565},{},{"id":568,"data":1262,"type":240,"tunes":1263},{"text":570},{},{"id":573,"data":1265,"type":240,"tunes":1266},{"text":575},{},{"id":578,"data":1268,"type":240,"tunes":1269},{"text":580},{},{"id":583,"data":1271,"type":42,"tunes":1272},{"text":585,"level":234},{},{"id":588,"data":1274,"type":240,"tunes":1275},{"text":590},{},{"id":593,"data":1277,"type":240,"tunes":1278},{"text":595},{},{"id":598,"data":1280,"type":291,"tunes":1281},{"code":600},{},{"id":603,"data":1283,"type":42,"tunes":1284},{"text":605,"level":234},{},{"id":608,"data":1286,"type":240,"tunes":1287},{"text":610},{},{"id":613,"data":1289,"type":240,"tunes":1290},{"text":615},{},{"id":618,"data":1292,"type":240,"tunes":1293},{"text":620},{},{"id":623,"data":1295,"type":42,"tunes":1296},{"text":625,"level":234},{},{"id":628,"data":1298,"type":240,"tunes":1299},{"text":630},{},{"id":633,"data":1301,"type":240,"tunes":1302},{"text":635},{},{"id":638,"data":1304,"type":42,"tunes":1305},{"text":640,"level":234},{},{"id":643,"data":1307,"type":240,"tunes":1308},{"text":645},{},{"id":648,"data":1310,"type":240,"tunes":1311},{"text":650},{},{"id":653,"data":1313,"type":291,"tunes":1314},{"code":655},{},{"id":658,"data":1316,"type":240,"tunes":1317},{"text":660},{},{"id":663,"data":1319,"type":42,"tunes":1320},{"text":665,"level":234},{},{"id":668,"data":1322,"type":240,"tunes":1323},{"text":670},{},{"id":673,"data":1325,"type":240,"tunes":1326},{"text":675},{},{"id":678,"data":1328,"type":240,"tunes":1329},{"text":680},{},"Post erfolgreich abgerufen",{"items":1332,"source":1353,"manualIds":1354,"manualMatchedIds":1355},[1333,1340,1346],{"id":1334,"slug":1335,"title":1336,"excerpt":1337,"featuredImage":1338,"publishedAt":1339},"445","qwen-3-6-in-production-release-runbook-ai-rollback-and-llmops-versioning","Qwen 3.6 u produkciji: Runbook za izdavanje, AI rollback i LLMOps verziranje","Qwen 3.6 nije samo još jedna nadogradnja modela. To je istovremeno događaj objavljivanja, scenario povratka na prethodnu verziju i problem verziranja. Ovaj članak objašnjava kako Qwen 3.6 treba tretirati u produkciji kroz LLMOps disciplinu, sledljivost promptova i modela, kontrolisano uvođenje i spremnost za povratak na prethodnu verziju zasnovanu na dokazima.","\u002Fuploads\u002F2026\u002F02\u002Fnew-qwen-3-5-plus-1771515512741-dcbi9p.webp","2026-05-04T02:49:00.000Z",{"id":1341,"slug":1342,"title":1343,"excerpt":1344,"featuredImage":1338,"publishedAt":1345},"384","new-qwen-3-5-plus","Novi Qwen 3.5-Plus: AI otvorenog koda je upravo postao ozbiljan.","Otkrijte revolucionarne funkcije i prednosti Alibabinog Qwen 3.5-Plus modela, AI otvorenog koda koji menja pravila igre za programere.","2026-02-19T10:23:00.000Z",{"id":1347,"slug":1348,"title":1349,"excerpt":1350,"featuredImage":1351,"publishedAt":1352},"486","source-of-truth-in-ai-systems-where-reliable-knowledge-actually-comes-from","Izvor istine u AI sistemima: Odakle pouzdano znanje zaista dolazi","Izvor istine definiše koji je izvor merodavan za određenu činjenicu ili stanje. Saznajte kako se razlikuje od RAG-a, porekla, memorije, konteksta, vektorskih baza podataka i sistema evidencije.","\u002Fuploads\u002F2026\u002F10\u002Fsource-of-truth-in-ai-systems-where-reliable-knowledge-actually-comes-from-1791479103235-6bq9em.webp","2026-10-08T13:02:00.000Z","fallback",[],[]]